AUTHOR ACCEPTED MANUSCRIPT

FINAL PUBLICATION INFORMATION

On Egypt's de facto Integration in the International Financial Market

The definitive version of the text was subsequently published in

Middle East Development Journal, 7(2), 2015-09-29

Published by Taylor and Francis and found at http://dx.doi.org/10.1080/17938120.2015.1083219

THE FINAL PUBLISHED VERSION OF THIS MANUSCRIPT IS AVAILABLE ON THE PUBLISHER’S PLATFORM

This Author Accepted Manuscript is copyrighted by World Bank and published by Taylor and Francis. It is posted here by agreement between them. Changes resulting from the publishing process—such as editing, corrections, structural formatting, and other quality control mechanisms—may not be reflected in this version of the text.

You may download, copy, and distribute this Author Accepted Manuscript for noncommercial purposes. Your license is limited by the following restrictions:

(1) You may use this Author Accepted Manuscript for noncommercial purposes only under a CC BY-NC-ND 3.0 IGO license http://creativecommons.org/licenses/by-nc-nd/3.0/igo.

(2) The integrity of the work and identification of the author, copyright owner, and publisher must be preserved in any copy.

(3) You must attribute this Author Accepted Manuscript in the following format: This is an Author Accepted Manuscript by Alnashar, Sara B. On Egypt's de facto Integration in the International Financial Market © World Bank, published in the Middle East Development Journal7(2) 2015-09-29 CC BY-NC-ND 3.0 IGO http://creativecommons.org/licenses/by-nc-nd/3.0/igo http://dx.doi.org/10.1080/17938120.2015.1083219

© 2021 World Bank

On Egypt’s de facto integration in the international financial market

Sara B. Alnashar

The World Bank Group, Cairo, Egypt

Received 14 April 2015; accepted 23 July 2015

The author is very grateful to Professor Peter Montiel, Williams College, who was the supervisor of unpublished academic work upon which this paper builds. Thanks go to Professor Peter Pedroni, Williams College, and Professor Alaa El Shazly, Cairo University, for useful discussions on empirical issues, and to Professor Omneia Helmy, Egyptian Center for Economic Studies and Cairo University, as well as Dr. Magda Kandil, International Monetary Fund, for helpful remarks, and to Dr. Hoda Selim, Economic Research Forum, for sharing useful resources, and to Dr. Salwa El-Antary for important clarifications regarding the history of Egypt’s financial system, and to two anonymous referees for constructive comments and suggestions. Any errors remain the responsibility of the author. Comments and suggestions are welcome at [email protected]. I. INTRODUCTION Egypt has been gradually dismantling legal restrictions on the entry and exit of capital flows since the early 1990s. Since then, the economy has been experiencing intermittent episodes of capital inflows and outflows. Most remarkably, in the wake of the international financial crisis of late 2008, the capital and financial account of Egypt’s balance of payments was hit hard as inflows dropped drastically. Net portfolio investment in Egypt decreased by a staggering 570 percent to reach an outflow of more than $9.2 billion in 2008/09. This was reversed to achieve a net inflow in the following year, but once again returned to a net outflow of $2.6 billion in 2010/11 due to the uncertainty brought about by the January 25th revolution. On the one hand, the volatility that portfolio investments have been exhibiting may point to a high degree of international financial integration/capital mobility. But on the other hand, preliminary assessment of the rates of return on financial assets in Egypt shows that their behavior is secular from that of its foreign counterparts. That is, a large and varying differential exists between the rates of return on domestic and foreign financial assets, which is inconsistent with a high level of financial integration/capital mobility.

Given such contradicting observations, how well is the Egyptian economy de facto integrated in the world financial market? And does capital mobility complicate the actions of the Central Bank of Egypt in a way that would inhibit monetary autonomy (especially since the exchange rate has been quite stable in Egypt even after the announced floatation in January 2003)? Moreover, while Egypt has become de jure financially open, what explains the persistence of this differential between return rates on domestic and foreign financial assets?

Answering these questions is important to understanding how the macroeconomy works. The domestic policies’ effectiveness in changing aggregate demand is determined— among other factors—by the degree of the economy’s international financial integration. This notion could be better explained in light of the impossible trinity. A country that has a pegged exchange rate and a high degree of capital mobility will have no autonomous monetary policy. On the other hand, a country that has a floating exchange rate and a high degree of capital mobility will be able to conduct an autonomous monetary policy, but high capital mobility would imply bigger fluctuations in the exchange rate in response to monetary shocks (Mundell 1963; Fleming 1962; Dornbusch 1976). Further, the volatility in financial assets’ returns may generate inflows and profit opportunities to foreign investors (from arbitrage), but may be destabilizing to the economy (Balli, Basher and Rana 2014).

1 In light of the above, the objective of the paper is two-fold: first, to formally test whether Egypt has become de facto financially open after the steps taken towards the de jure liberalization of the capital and financial account of the balance of payments;1 second, to explain the presence of the large and varying rate of return differential between domestic and foreign financial assets.

For the first objective of this study, we ran two empirical tests: the uncovered interest parity (UIP) test and a monetary autonomy test. Using monthly data, the UIP failed to hold, as the exchange rate-adjusted differential between Egyptian 3-month Treasury bill rates and the US counterpart has been found to be non stationary for the whole period under study (January 2000-December 2011) and for the shorter period that witnessed large capital inflows (July 2004–June 2008). Also, the monetary autonomy test has shown that the growth rate in the monetary aggregate (M2) Granger-causes movements in the exchange rate-adjusted differential for the whole period under study, as well as the shorter period of high capital inflows. This means that despite the stable exchange rate, and the de jure capital and financial openness, the Central Bank of Egypt has been capable of stimulating changes in the domestic interest rate. This means that monetary autonomy has been preserved. The UIP and monetary autonomy tests both point to imperfect international financial integration for Egypt.

For the second objective of this study, initially we test for the presence of a long-run equilibrium relationship between the interest rate differential and its potential macro determinants. For the period 2001/02–2010/11, using quarterly data, cointegration was detected between the following five variables: (1) nominal interest rate spread between Egyptian and US 3-month Treasury bill rates, (2) log of the bilateral exchange rate (LE/$), (3) differential between logs of monetary aggregates (M2) in Egypt and the US, (4) differential between the logs of real output in Egypt and the US, and (5) expected differential between the inflation rates in Egypt and the US. Having detected a long-run equilibrium relationship, a vector error-correction model (VECM) was estimated. The generated forecast error variance decomposition shows that the expected inflation differential is the most important contributor to the variation in the differential between domestic and foreign 3-month Treasury bill rates.

To the best of our knowledge, international financial integration has not been

1 A country’s de facto financial integration may deviate from its de jure status. According to Bekaert and Harvey (1995), markets may turn out to be more financially integrated than one might expect based on prior knowledge of capital restrictions, whereas other markets seem segmented even though their capital markets are accessible to foreign investors. 2 investigated empirically for the Egyptian case, specifically in the context of macroeconomic management. Thus, this paper contributes to understanding one of the important elements of Egypt’s macroeconomic environment, through:

• Empirically quantifying the de facto degree of Egypt’s international capital mobility (financial integration). In doing so, we also gauge the extent to which monetary autonomy is maintained, in the face of a stable exchange rate.

• Measuring the relative importance of macro determinants affecting the degree of Egypt’s international financial integration.

Empirically quantifying Egypt’s degree of de facto capital mobility is important in designing macro policies; understanding how these policies succeed in affecting aggregate demand, and clarifying to what extent de jure (versus de facto) capital openness has implications for the effectiveness of macro policies.

This paper is organized as follows: After the introduction in Section I above, Section II provides an overview of the main areas of the Egyptian economy that are relevant to the study of international financial integration: (1) the gradual de jure liberalization of the capital and financial account of Egypt's balance of payments, and (2) the evolution of the domestic interest rate in Egypt, as well as that of the US. Thus the purpose of this section is to give a picture of the macroeconomic context that guides our empirical investigation. Section III is dedicated to the literature review on the approaches to measuring de facto international financial integration that are later applied in the empirical section. Section IV presents empirical findings of this study based on the UIP test, the test of monetary autonomy, and the cointegration/VECM analysis. Section V concludes.

II. DEVELOPMENTS IN EGYPT'S CAPITAL AND FINANCIAL ACCOUNT, AND DOMESTIC FINANCIAL MARKET SINCE THE EARLY 1990S In this section, we trace the developments of the main variables that pertain to the analysis of Egypt’s integration in the world financial market. This helps in getting a preliminary assessment of the issue before embarking on the more formal testing, and also serves as a

3 basis upon which necessary assumptions will be built in the empirical section, and for identifying the time period that will be most suitable for the empirical tests.

We start by tracking capital and financial liberalization since Egypt adopted the Economic Reform and Structural Adjustment Program (ERSAP) in sub-section (II.1). This will shed light on the openness of Egypt’s capital and financial account de jure. We then cover the evolution of domestic interest rates in Egypt, against their foreign counterparts in sub-section (II.2), and this will show initial signs of Egypt’s capital and financial openness de facto.

II.1. Egypt's de jure financial openness

Egypt has started taking steps towards de jure financial openness since the early 1990s under ERSAP. This could be illustrated by a plot of the Chinn and Ito (2008) “capital openness index” that takes on higher values the more open the country is to cross-border capital transactions (Figure 1).2 The details of the developments in lifting capital controls and significant milestones are outlined in Appendix (1)

Figure 1. Chinn and Ito’s Capital Account Openness Index (1970–2010)

Source: Chinn and Ito index online http://web.pdx.edu/~ito/Chinn-Ito_website.htm.

Figure 1 shows that Egypt started to gradually liberalize its capital and financial account during the early 1990s. As per the Chinn-Ito index, Egypt’s de jure capital and financial

2 The Capital Openness Index is based on the binary dummy variables that codify the tabulation of restrictions on cross-border financial transactions reported in the IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions (AREAER). The index is the aggregation of the dummy variables that are assigned to the four major categories of the restrictions on external accounts (1) multiple exchange rates; (2) restrictions on current account transactions; (3) restrictions on capital account transactions; and (4) surrender requirements of export proceeds. Starting 1996, the IMF switched from binary coding to a disaggregated coding. The lowest aggregate score reflects presence of restrictions in all categories of restrictions on external accounts, and vice versa (Chinn and Ito 2008). 4 openness continued rather steadily until the observed drop in the index in 2003. This drop is explained by several restrictive measures temporarily introduced to accompany the announcement of floating the Egyptian pound in January 2003.3 Those measures were later removed in 2004.

With the exception of the year 2003, Egypt was classified as having a de jure open capital and financial account of the balance of payments, according to the Chinn-Ito index during the period 2001–2008. The years 2009 and 2010 have seen declines in the level of de jure openness as the economy took measures to hedge the adverse effects of the global financial crisis. In fact, Egypt’s score on this index reached “2.45573,” which is the highest possible score, and is comparable to that of industrial countries such as the US, the UK and a few emerging markets (Chile and Jordan). However, most other emerging economies have maintained lower scores on this index during 2004-2008, such as Brazil (a score of 0.26), South Africa (-1.16), India (-1.16) and Indonesia (1.13).

The legal liberalization of the capital and financial account of the balance of payments seems to have affected the volume of capital inflows in Egypt. We are particularly interested in portfolio flows as they are most relevant to the analysis of Egypt’s international financial integration (Figure 2).

Figure 2. Capital and Financial Account of the Balance of Payments and Net Portfolio Inflows (% of GDP) (1990/91–2010/11) 6

4

2 Portfolio 0 inflow

-2 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

(% of GDP) Capital and -4 Financial Account -6

-8 Source: Central Bank of Egypt website, online time series: http://www.cbe.org.eg/timeSeries.htm.

3 In 2003, a special exchange rate (of LE 5.35 per $1) applicable to key imported foods was introduced. Also, private and state-owned exporting companies were required to sell at least 75 percent of their foreign currency earnings to state-owned banks. Foreign investors’ remittances of profits and dividends were made subject to delays. 5 Portfolio inflows were quite modest during the early 1990s, but started to pick up during the early 2000s, especially after 2003/04 coinciding with the de jure liberalization of the capital and financial account of the balance of payments. In 2005/06, portfolio inflows reached 3 percent of GDP. But that was reversed into a portfolio outflow (reaching a volume of $9.2 billion, amounting to 5 percent of GDP in 2008/09) on account of the international financial crisis. This was switched once again in 2009/10 to achieve an inflow of 4 percent of GDP in 2009/10 when the Egyptian economy started recovering from the crisis. Thus it seems that the period 2003/04–2007/08 is of particular interest, due to the increased activity in portfolio investments. This period also coincides with the period of de jure capital and financial openness according to the Chinn-Ito index discussed above. The empirical part of this study will attempt to investigate Egypt’s de facto financial openness during this period in particular.

II.2. Egypt's de facto financial openness

We now turn to analyzing the behavior of the domestic short-term interest rate in Egypt against those of foreign counterparts. This will help us assess to what extent the international financial market influences the behavior of domestic interest rates in Egypt.

In Figure 3, we trace the developments in the 3-month Treasury bill rate4 against that of the US counterpart, along with the exchange rate-adjusted differential between the two rates. It shows that fluctuations in the Egyptian 3-month Treasury bill rate are independent from those in the US counterpart, with the exception of the short period January 2006–March 2008, during which there is a noticeable correspondence between the two series, as well as a decline in the exchange rate-adjusted differential between them. But apart from that period, the exchange rate-adjusted differential seems to move in parallel to the Egyptian 3-month Treasury bill rate. This may be held as preliminary evidence that the Egyptian interest rates are not strongly influenced by the foreign financial market.

4 Treasury bills were first introduced in 1991 on a weekly auction basis. The objective was to initiate a market mechanism to determine interest rates, introduce an instrument to regulate banks' reserves, and absorb excess liquidity as well as dampen the impact of capital inflows (i.e., sterilization ), besides financing the budget deficit (El-Refaie 2002; Al-Mashat and Billmeier 2007). 6 Figure 3. Three-Month Treasury Bill Rates in Egypt and the US, and Exchange Rate-Adjusted Differential (July 1996–December 2011) (in percent, annually)

16 14 12 10 8 6 4 2 0 Jun-96 Jun-97 Jun-98 Jun-99 Jun-00 Jun-01 Jun-02 Jun-03 Jun-04 Jun-05 Jun-06 Jun-07 Jun-08 Jun-09 Jun-10 Jun-11 Dec-96 Dec-97 Dec-98 Dec-99 Dec-00 Dec-01 Dec-02 Dec-03 Dec-04 Dec-05 Dec-06 Dec-07 Dec-08 Dec-09 Dec-10 Dec-11

Egy 91-day T-bill rate US 13-week T-bill rate Exchange Rate-Adj. Differential

Source: Prior to 2003, monthly data on the Egyptian 3-month Treasury bill rate were obtained from Central Bank of Egypt upon request. Starting 2003, data were obtained from the Central Bank of Egypt, Monthly Statistical Bulletin, various issues. The US 13-week Treasury bill rate was obtained from the International Financial Statistics online database. * Note: Exchange rate-adjusted differential is calculated by author according to the following formula: (1 + it) - (1 + it ) St+k/St. where i and i* are the domestic and foreign interest rates, respectively. S is the nominal exchange rate (LE/USD), and is obtained from the International Financial Statistics online database. k is the maturity period, equal to 3, in this case.

The three-month Treasury bill rate (along with the other short-term domestic interest rates) remained elevated5 and almost constant from 1996/976 up until 2000/01, after that it started to move more freely. This behavior could be explained by the fact that up until the year 2000/01, Treasury bills represented the main placement for banks' excess liquidity (i.e., funds which were not given out as bank loans nor invested in projects were used to purchase Treasury bills); noting that “excess reserves” were the main operational target for monetary policy. El-Refaie (2002) explained that the Central Bank of Egypt chose to keep the interest rate on Treasury bills constant in an attempt to maintain a balance between conflicting policies: On one hand, there was a need to issue more Treasury bills to finance the budget deficit and to absorb excess liquidity caused by capital inflows (which meant accepting higher interest rates on Treasury bills), while on the other hand, there were growth targets (which necessitated keeping interest rates low). Another factor that contributed to the stable interest rate on Treasury bills prior to 2001 was the dominance of the state in the banking sector, which tended to create rigidities in the interest rate structure (Al-Mashat and Billmeier 2007).

5 It is worth noting that domestic interest rates shot upwards following the financial liberalization reforms that were undertaken in the early 1990s (Abdel-Khalek 2001). Also, the Central Bank of Egypt maintained a tight monetary policy under ERSAP, and thus there was a positive and noticeable differential between Egyptian and developed countries' interest rates (El-Refaie 2002). 6 Monthly data for the 3-month Egyptian Treasury bill rate are available from the Central Bank of Egypt (upon request) starting from fiscal year 1996/97. 7 The larger flexibility of the 3-month Treasury bill rate starting from 2000/01 could be attributed to several factors. The overnight domestic currency interbank market was introduced in 2001. Interbank lending therefore became an alternative placement for banks' excess liquidity. The Central Bank of Egypt gradually introduced other supportive monetary policy tools. These included repos, reverse repos and short-term deposits at the Central Bank of Egypt. The introduction of the interbank market and the new tools for conducting monetary policy enhanced the degree of market determination of short term interest rates, such as the Treasury bill rate. Finally, in 2005 the Central Bank of Egypt’s monetary policy framework was restructured. The corridor system was introduced and the overnight interest rate on interbank transactions was formally adopted as the main operational target instead of the excess bank reserves (Al-Mashat and Billmeier 2007; Moursi, El-Mossallamy and Zakareya 2007).

Moreover, the fluctuations that are observed in the Treasury bill rate since 2000/01 coincided with developments in the exchange rate; that is, the step devaluations (later on depreciation) of the Egyptian pound. Based on an empirical study that covered the period 2000/01–2007/08, Selim (2012a) provides evidence that the monetary authority had been strongly reacting to changes in the exchange rate.

In light of the above analysis, we decide to start the empirical analysis from 1999/2000 as it marks the beginning of de jure capital and financial openness, according to the Chinn-Ito index. This year also marks the beginning of the enhanced variability in the exchange rate- adjusted differential between the Egyptian and US 3-month Treasury bill rates. Moreover, the Egyptian interest rates prior to 2000 could not be described as “market-determined” (as it was constant) and so would not be relevant to the analysis of Egypt’s international financial integration. Further, given that the exchange rate has been closely managed even after the announced floatation in 2003 (Selim 2012b), thus exploring the extent of monetary autonomy will have implications for the degree of Egypt’s financial integration. In other words, the stable exchange rate will allow us to interpret the preservation of monetary autonomy as a low degree of financial integration, by virtue of the impossible trinity.

8 III. LITERATURE REVIEW: HOW DO WE MEASURE DE FACTO INTERNATIONAL FINANCIAL INTEGRATION? The literature has presented several definitions of international capital mobility/ financial integration, upon which the empirical investigations are usually based. The most popular definitions include the following four, presented in ascending order of specificity (Frankel, 1991): (1) The Feldstein-Horioka definition: A small economy is deemed open to international capital flows when changes in the domestic savings rate have no effect on the domestic investment rate. As such, “exogenous changes in domestic saving (that is, in either private savings or government budgets) can be easily financed by borrowing/lending abroad, and thus need not crowd out/in domestic investment”. (2) Real interest parity: refers to the case when arbitrage (international capital flows) leads to the equalization of expected real returns across similar domestic and foreign assets. (3) Uncovered interest parity: When arbitrage leads to the equalization of expected returns on similar domestic and foreign financial assets (the respective countries’ bonds). (4) Covered (closed) interest parity: When arbitrage equalizes return rates, provided that exchange rate risk is hedged by forward cover.7

Out of the definitions above, the uncovered interest parity is arguably the most relevant/applicable to the Egyptian case: The Feldstein-Horioka definition (Investment- Savings correlations) is the strictest one, as it requires real interest parity to also hold. Real interest parity is quite strict as well, because it not only requires uncovered interest parity to hold, but also that the expected real depreciation is zero. Further, real interest parity incorporates both financial and real integration. The covered interest parity condition cannot be tested for most emerging markets, such as Egypt, due to the absence of forward markets for these countries’ currencies.

In addition to the empirical tests that follow from the definitions of international capital mobility (financial integration), other tests were developed based on the impossible trinity hypothesis, namely the monetary autonomy tests. Under a pegged exchange rate, measuring

7 “Asset pricing” can be considered another strand of literature on the study of international financial market integration. According to Bekaert and Harvey (1995), asset pricing models assume that capital markets are either totally segmented, or perfectly integrated, or in between the two poles. “Capital Asset Pricing Models (CAPM)” are tested in case the market is assumed to be segmented. “World CAPM”, “CAPM with exchange rate risk”, and the “World Arbitrage Pricing Theory” are among the models tested in case the market is assumed to be integrated. Finally, the “Mild segmentation model” assumes that the market is partially integrated/segmented. These models have limited applicability when trying to assess international financial integration in the context of macroeconomic management. 9 the degree of monetary autonomy can be viewed as the flip-side of international capital mobility. A country that retains monetary autonomy, with a stable exchange rate, must be imperfectly integrated in the world financial market. Egypt has maintained a pegged exchange rate up until the early 2000s, and empirical work (Selim 2012b) has also proved that the exchange rate continued to be closely managed even after the announced floatation in January 2003. Thus, testing for the degree of monetary autonomy for the case of Egypt will have implications for the degree of the country’s international financial integration.

In what follows, we focus on the two methodologies that are applied later in the 8 empirical section of this paper: The Uncovered Interest Parity and the Monetary Autonomy.

III.1. Uncovered Interest Parity (UIP)

This approach assesses the responsiveness of domestic interest rates to changes in foreign interest rates, and to what extent changes in the relative rates of return on domestic and foreign financial assets give rise to cross-border arbitrage flows (Montiel1994). Interest rate differentials are thus used to assess the degree of cross-border capital mobility arising from the equalization of return rates on financial assets in two different countries. The argument is that, if capital is perfectly mobile, then its rate of return should be equal across countries; and hence, no interest rate differential should exist (Hussein and de Mello 1999).

In the context of measuring financial integration, the UIP condition is formally stated as:

* 9 1 + it = Et [(1 + it ) St+k/St ] (1)

* where it is the domestic interest rate, it is the foreign interest rate, and St and St+k are the domestic-currency price of foreign exchange in the current period and the next, respectively.

8 See Montiel (1994) and Frankel (1996) for a review of other popular methodologies for empirically measuring international financial integration. 9Equation 1 is an alternative representation of the UIP condition. Another representation is as follows:

St+1-St = a + b (it – it*)+ et+1, in which case the UIP condition would consist of testing the joint hypothesis that a=0 and b=1 (Chinn and Meredith 2004).

However, we use the specification derived from Equation 1 above, as we are interested in the responsiveness of the domestic interest rate against changes in the exchange rate-adjusted foreign counterpart.

Further, for the Egyptian case, it is shown in Table (1) below that the term (St+1-St) is stationary, whereas the term (it – it*) is non-stationary, therefore the UIP condition cannot be tested in the fashion just described here. 10 The next period (t+k) is determined by the maturity of the financial asset whose return rate we 10 are interested in. Et is the expectation operator. The expected exchange rate however is subjective, since it is unobservable. Thus, when applying the uncovered interest parity test, * “rational expectations” is imposed in order to allow the use of ex-post values of [(1 + it )

St+k/St] as equivalent to their expected values plus a prediction error ‘e’. As such,

* * Et [(1 + it ) St+k/St ] = [(1 + it ) St+k/St ] + e (2) where the prediction error ‘e’ should be a mean-zero, serially uncorrelated random variable. In other words, the ex-post interest rate differential (which is equal to the prediction error, e) would be stated as follows:

* Exchange rate-adjusted differential = dt = (1 + it) - (1 + it ) St+k/St (3)

Provided that all variables that appear in equation 3 are stationary, the UIP condition consists of testing the null hypothesis that dt has a zero mean (i.e., no differential exists) and is serially uncorrelated (that is, successive lags of the differential /prediction error should not be correlated).

* However, if the terms (1+it) and [(1 + it ) St+k/St] are non-stationary, then for the uncovered interest parity condition to hold, the exchange rate-adjusted differential dt should be “stationary.” In other words, a long-run relationship should exist between the return rates on the domestic and exchange rate-adjusted foreign financial assets. Empirically, this means that they must be co-integrated, such that the exchange rate-adjusted differential between the two rates is stationary.

Obstfeld, Shambaugh and Taylor (2005) described the UIP condition as explained above: That is, the nominal international interest rate differential being composed of a statistically stationary expected depreciation, and a stationary risk premium.

To the best of our knowledge, the UIP has not been tested in the fashion described above to try to gauge Egypt’s integration in the world financial market. Thus we present here an example of how the UIP condition was employed for the People’s Republic of China (PRC). Cheung, Chinn and Fujii (2006) employ the UIP condition to test for the PRC’s financial integration with each of the following economies: Hong Kong, Taiwan, Japan and

10 For example, if we are considering the 3-month Treasury bill rate, then k is equal to 3. 11 the US.11 Specifically, for the period February 1996–June 2002, unit root tests were applied on the uncovered interest differential between domestic and exchange rate-adjusted foreign one-month interbank interest rates for the following economy-pairs: PRC/Hong Kong, PRC/Taiwan, PRC/US and PRC/Japan. For all pairs—except the PRC/US—the null hypothesis of presence of a unit root was strongly rejected. When a dummy variable for the 1997 financial crisis was included, they were also able to reject the null hypothesis of the presence of a unit root for the PRC/US series. Thus, after accounting for the 1997 financial crisis, all the uncovered interest differential series were proven to be stationary and shocks to the UIP were deemed as transitory. Based on this test, Cheung, Chinn and Fujii concluded that PRC is strongly integrated with the other economies of Greater China (Taiwan and Hong Kong) as well as its major trading partners (Japan and the US). This work has been held as evidence of the fading effectiveness of de jure capital controls in China.

The UIP test has been applied widely in the context of investigating financial integration. But the failure thereof has been interpreted with caution. While departure from UIP may be an indicator of a low degree of financial integration, it should be noted that this is a common empirical finding, and may not be enough to prove that a country is not well integrated in the world financial market. That is because for the UIP to hold, two stringent conditions need to be met: (1) investors need to be “rational” with regards to expectations of the future spot exchange rate, and (2) investors need to be risk-neutral, such that no premium exists between domestic and foreign interest rates. In reality, those two conditions are very difficult to hold, and literature has dubbed them as the textbook/traditional reasons for departure from the UIP: expectations are not necessarily rational, and investors usually demand a premium as they are risk-averse. This risk premium may be time-variant due to exchange rate risks, political risks, etc., which thus leads to failure of the UIP.

But in addition to these traditional reasons, Alper, Ardic and Fendoglu (2009) also state other reasons for the unfavorable empirical evidence for the UIP condition, including: existence of transaction costs, possible effects of central bank interventions (notably to stem exchange rate fluctuations in the “fear-to-float” context), and the possibility that investors may care for real rather than nominal returns.

11 Cheung, Chinn and Fujii (2006) also test for other criteria of integration, namely, real interest parity and real purchasing power parity. But we focus solely on their findings in the UIP test as they are the most relevant to this study. 12 III.2. Test of Monetary Autonomy

Under fixed exchange rates, tests of monetary autonomy can be based on whether the central bank's monetary policy tools are successful in changing the stock of money and domestic interest rates. As discussed earlier in the context of the impossible trinity, if a country maintains a pegged exchange rate regime, then a high degree of capital mobility will render the monetary policy ineffective in stimulating changes in domestic monetary aggregates, and ultimately in aggregate demand.

For the central bank to change the stock of money, it engages in open market operations in order to initiate a change in its domestic assets. The change in the domestic assets of the central bank will in turn lead to a change in domestic interest rates. In a financially closed economy, this would result in a change in the stock of money. However, if a country is characterized by perfect capital mobility, then the change in domestic interest rates will soon be eliminated through arbitrage flows. The change in the central bank’s domestic assets (that took place because of open market operations) will be offset by a change in its foreign assets, while the stock of money remains unchanged (Montiel 1994). More specifically, under perfect capital mobility, a credit expansion will create an equivalent capital outflow, and vice versa (Rennhack and Mondino 1988).

This property of capital mobility has been formally tested by constructing a structural model of the financial sector of an economy, and estimating an offset coefficient. This coefficient gives the amount of capital outflow per unit of expansion of domestic credit.12 The offset coefficient ranges from zero to -1. A value of -1 would indicate a high degree of capital mobility (Montiel 1994; Rennhack and Mondino 1988). However, such approaches that rely on the extent to which capital openness offsets domestic credit expansion are problematic, due to the difficulty of identifying exogenous credit shocks (Obstfeld, Shambaugh and Taylor 2005).

Obstfeld, Shambaugh and Taylor assess the validity of the impossible trinity using data series over 130-years long. In an attempt to capture the monetary independence element of the trinity – under assumptions of perfect capital mobility and pegged exchange rates – they relied on the extent to which local interest rates diverge from the world interest rate (that is, the interest rate in some well-defined base-country market).13 In their tests to validate the

12 Unit of expansion here means: expansion of domestic credit by one unit of domestic currency. 13 They argued that “[e]ven if the interest rate is not the primary instrument of monetary policy, it should be directly affected by monetary policy changes, and thus would still serve as a measure of the stance of policy. If the 13

impossible trinity, they estimated the following equation.

= + + (4)

𝑖𝑖𝑖𝑖 𝑏𝑏𝑏𝑏𝑏𝑏 𝑖𝑖𝑖𝑖 where is the local interest∆𝑅𝑅 rate;𝛼𝛼 𝛽𝛽 ∆is𝑅𝑅 the foreign𝑢𝑢 interest rate.

If a country∆𝑅𝑅𝑖𝑖𝑡𝑡 has a permanently and credibly∆𝑅𝑅𝑏𝑏𝑖𝑖𝑡𝑡 pegged exchange rate, zero monetary autonomy would be concluded when β = 1, but only under perfect capital mobility. Similarly, Aizenman, Chinn and Ito (2010) define the extent of monetary independence as "the reciprocal of the annual correlation of monthly money market interest rate in the home country and base country". Higher values of the index indicate a larger degree of monetary independence. They calculate this index for 171 countries, of which 27 are emerging markets, including Egypt. Egypt is found to have moderate monetary independence, with an index of 0.50 in the year 2007, at a time when Egypt enjoyed relative exchange rate stability, and a de jure open capital account, according to the indices compiled by the authors. In contrast, Chile for instance, had a higher monetary independence index of 0.96 in 2007, despite a comparable level of de jure capital openness as that of Egypt; but that is because Chile has a less stable exchange rate. On the other hand, Argentina is found to have a higher level of monetary independence, compared to Egypt, with a score of 0.74 in 2007, even though the country has a stable exchange rate; but that is under a lower degree of de jure capital openness, as reported by the authors.

Alternatively, Montiel (1989) tried to capture monetary autonomy by Granger-causality tests. Under a pegged exchange rate, and assumed perfect capital mobility, the domestic financial aggregates, such as money (M1, M2) or domestic credit, should not Granger-cause movements in domestic interest rates, nominal or real output. He presented empirical

interest rate is insulated from global market conditions by capital controls, this is important as well in that it demonstrates how capital controls can allow monetary autonomy and a fixed exchange rate to exist simultaneously.” 14 evidence for 12 developing countries for the period 1962-1986, showing that past changes in domestic financial aggregates (money supply or domestic credit) helped predict nominal GDP. Given that the countries were characterized by a fixed exchange rate, the Granger- causality tests were used as evidence to conclude that capital mobility/financial integration was imperfect for those countries.

We turn next to the application of these two tests to assess Egypt’s de facto financial integration. We first test whether Egypt has become de facto open financially, upon the de jure capital and financial liberalization. As the tests show that Egypt’s de facto financial integration is limited, we try to explain why that is the case. So, we try to assess the reasons behind the persistence of the large and time-varying interest rate differential between domestic and financial assets. This is done by employing the Dornbusch model in the context of a cointegration/vector error-correction analysis in order to identify the relatively more important contributors to the variability that the interest rate differential exhibits.

IV. THE EMPIRICS: ON EGYPT’S DE FACTO INTEGRATION IN THE WORLD FINANCIAL MARKET The objective of this empirical section is: First, to formally test how responsive domestic interest rates are to foreign counterparts. This is tested using the uncovered interest parity condition as well as the monetary autonomy test. Second, to model the behavior of that interest rate spread that exists between domestic and foreign financial assets against the macro variables proposed by the fundamentals-based Dornbusch approach. This is done through 14 cointegration/vector error-correction analysis.

In sub-section IV.1, the UIP condition is tested. It is shown that the exchange rate- adjusted differential between Egyptian and US interest rates is non-stationary, mainly due to the non-stationarity of the interest rate spread between the two countries’ financial assets, whereas the exchange rate depreciation is found to be stationary. This brings us to sub-section IV.2 in which the monetary autonomy test is applied, using Granger causality. Given the stationarity of the exchange rate depreciation (as validated in the previous test), the effectiveness of monetary policy helps us make an inference regarding the degree of de facto international financial integration. In other words, retaining monetary autonomy under a stable exchange rate is the other side of the coin of a low degree of international financial

14 Discussion of data sources and issues is deferred to Appendix 2. 15 integration. Finally, in sub-section IV.3, we will assess the macro determinants of the time- varying spread between domestic and foreign interest rates in order to explain why Egypt’s international financial integration continues to be limited, despite dismantling legal restrictions on international capital flows. After estimating a Vector Error-Correction Model, it is shown that all variables in the model hold predictive content for the “interest spread”; the variable of interest. Then, variance decompositions are calculated in order to determine the relatively more important contributors to the variability in the interest rate spread.

IV.1. . The Uncovered Interest Parity (UIP) Test

For UIP to hold with the presence of a persistent interest rate differential, the exchange rate-adjusted interest rate differential should be stationary. We have presented the formula of the exchange rate-adjusted interest rate differential in the literature review section, but we reiterate it here for convenience:

* Ex/Rate-Adj. Differentialt= dt = (1 + it) - (1 + it ) St+k/St (3 revisited)

Where i and i* are the interest rates on domestic and foreign financial assets (Egyptian and

US 3-month Treasury bill rates) respectively, and St and St+k are the nominal exchange rate in the current period and at the end of the maturity period, respectively. The nominal exchange rate here is measured as the domestic-currency price of foreign exchange. So for the bilateral exchange rate between Egyptian and US currencies, “S” is expressed in Egyptian pounds per 1 US dollar.

Using monthly data, the time series properties of the exchange rate-adjusted differential are investigated. The results of the Augmented Dickey Fuller (ADF) test are reported in Table 1. As shown, the exchange rate-adjusted interest rate differential is non-stationary, when tested either for the whole dataset (January 2000–September 2011), or for the short period that witnessed an episode of large portfolio inflows in Egypt (July 2004–June 2007). This implies a failure of UIP.

16 Table 1. Augmented Dickey-Fuller Test Results for Terms Included in the UIP Condition

Stationarity test details for variable in Stationarity test details for variable in LEVEL FIRST DIFFERENCES

MacKinnon MacKinnon one-sided p- one-sided p- Order of Series tested Time period; values Time period; ADF values ADF test ADF test ADF test integration for # of included associated # of included test associated details statistic details of variable stationarity observations with t- observations statistic with t- in level statistic of statistic of the ADF test the ADF test

2000:03 – 2000:03 –

Exchange 2011:09; Constant, 2011:09; Constant, -1.668 0.448 -9.422 0 rate- 1 lag 0 lags adjusted 139 obs 139 obs Egy-US 3- I(1) month T-bill 2004:07 – 2004:07 – rate 2008:06; Constant, 2008:06; Constant, -2.089 0.249 -5.240 0 differential 1 lag 0 lags 48 obs 48 obs

2000:03 – 2000:03 –

2011:12; Constant, 2011:12; Constant, Interest rate -1.465 0.548 -9.702 0 1 lag 0 lags spread between Egy 142 obs 142 obs I(1) and US 3- month T-bill 2004:07 – 2004:07 – 2008:06; Constant, 2008:06; Constant, rates -2.11 0.241 -4.921 0 1 lag 0 lags 48 obs 48 obs

2000:04 – 2011:09; Constant, -5.362 0.00 2 lags Exchange 138 obs rate I(0) depreciation 2004:07 – 2008:06; Constant, -5.091 0.00 0 lags 48 obs

It is useful to disentangle the components of the exchange rate-adjusted interest rate differential in order to investigate the time series properties of each component, and identify which term contributes to its non-stationarity. It appears that the spread between the Egyptian and US 3-month T-bill rates is responsible for the non-stationarity of the exchange rate-

17 adjusted interest rate differential.15 Unsurprisingly, exchange rate depreciation is stationary, and this is because the nominal exchange rate in Egypt is closely managed, even after the 16 announcement of the de jure floatation in January 2003.

As pointed out in the literature review section, failure of the UIP may only be a preliminary (not decisive) indicator of the degree of international financial integration. Therefore, before we move to the explanation of the non-stationarity of the interest rate differential between domestic and US financial assets, we turn to the test of monetary autonomy as another empirical test for Egypt’s de facto international financial integration.

IV.2. . Monetary Autonomy Test

Are the actions of the Central Bank of Egypt actually succeeding to stimulate changes in monetary variables (such as the domestic interest rate)? Or are the actions of the Central Bank of Egypt frustrated by capital mobility? In other words, examining the extent of monetary autonomy in Egypt will help us draw conclusions on the degree of Egypt’s financial integration, given that the exchange rate is quite stable (i.e., the exchange rate is not flexible; and so does not fluctuate to absorb nominal shocks). Monetary autonomy would not be preserved in such a setting, unless financial integration/capital mobility is imperfect.

We approach this question of monetary autonomy through running a Granger-causality test that investigates whether changes in money supply bear predictive content for the exchange rate-adjusted differential between domestic and US 3-month Treasury bill rates. If changes in the domestic monetary aggregate Granger-cause movements in that differential, monetary policy is said to be autonomous, as it creates deviations from uncovered interest parity. Given this finding—coupled with Egypt’s stable exchange rate—we could conclude that Egypt’s financial integration is imperfect. The Granger-causality test was run twice: the first time, for the sample as a whole: (January 2000–September 2011), and the second time, for the period that had registered the highest score in terms of de jure capital and financial openness (July 2004–June 2008). The lag structure of the Granger-causality tests was chosen

15 This piece of information is useful for the cointegration/VECM analysis that is conducted in this study (the third empirical test), as the spread between the Egyptian and US 3-month Treasury bill rate (i – i*) enter the cointegration /VECM as the variable of interest, in the investigation of the reason behind the low level of Egypt’s international financial integration. 16 Selim (2012b) provides empirical evidence that Egypt’s exchange rate has been de facto fixed years after the announced floatation. 18 by running unrestricted 2-variable vector-autoregressions (VAR) consisting of the endogenous variables in their levels (logM2 and the exchange rate-adjusted differential between Egy-US Treasury bill rates). According to the Akaike information criterion for the estimated VAR in levels, two lags were selected for the Granger-causality test that was run for the sample as a whole, but four lags were selected for the shorter period that witnessed a surge in Egypt’s de jure capital and financial openness. Results are presented in the following table.

Table 2. Granger-Causality Test Results Null Hypothesis: ∆(LM2) does not Granger Cause ∆(Ex/Rate-Adj. Differential)

No. of lags No. of Obs. F-Statistic Prob. Conclusion chosen by Akaike info. criterion

Reject Ho at 10 percent 2 138 2.880 0.0597 Full sample (2000M1-2011M9) signif.

Reject Ho at 10 percent 4 48 2.363 0.0698 Shorter sample (2004M7-2008M6) signif.

As displayed in Table 2 above, we are able to reject the null hypothesis of no Granger- causality for the entire sample period (January 2000–September 2009) as well as for the shorter period (July 2004–June 2008) that witnessed an increase in financial flows on the back of Egypt’s de jure financial openness. That is, the growth of M2 helped predict changes in the exchange rate-adjusted interest rate differential for the entire period under study. This refutes the argument that monetary autonomy is challenged by increased financial openness in Egypt.17 The bottom-line is that during the periods under investigation, Egypt has enjoyed de jure openness, but an imperfect degree of de facto financial integration.

Now our next task is to try to identify the reasons behind this limited degree of de facto financial integration. We look into this question by identifying the contributors to the variation in the non-stationary interest rate spread between Egyptian and US financial assets. Thus the objective of the following test is to explain the behavior of the interest rate spread against its potential macro-determinants.

17 That the “conduct of monetary policy is complicated by capital mobility” is a recently oft-cited argument (for example in: IMF, 2009 and 2010 Article IV consultation, p. 17 and p. 5, respectively) 19 IV.3. . Cointegration/VECM Analysis

In the two preceding empirical tests, we have seen that the interest rate spread18 is not entirely explained by the expected depreciation in the exchange rate (i.e., the UIP fails to hold). We have also seen that monetary autonomy is preserved despite the stable exchange rate.

Now the objective of our VECM analysis is to explain the variation in the interest rate spread in an attempt to find the reasons behind Egypt’s low de facto financial integration. This is done in the context of the “sticky price monetary approach” introduced by Dornbusch (1976). The interest rate spread (i – i*) is central to the mechanism that the Dornbusch model19 presents. According to Dornbusch, a contractionary monetary policy raises the interest rate spread, attracts capital inflows and leads to nominal exchange rate appreciation such that it may overshoot its long-run equilibrium value. While the nominal exchange rate constituted the focus of the Dornbusch model, we may also make use of this model to explain the behavior of the interest rate spread. This is done by running a VECM where all the variables that appear in the Dornbusch model are allowed to be endogenous so that we can distill the variance decompositions of the interest rate spread as generated from the VECM analysis. Thus the Dornbusch approach provides a reference point for the model specification.

So our VECM consists of the following 5 endogenous variables: 1. St: log of the bilateral * exchange rate (LE/USD), 2. (it–it ): nominal interest rate spread between Egyptian and US 3- * month Treasury bill rates, 3. (Mt – Mt ): differential between logs of monetary aggregates

(M2) in Egypt and the US, 4. (Yt–Yt*): differential between the logs of real output in Egypt e e* and the US, and 5. (πt – πt ): expected differential between the inflation rates in Egypt and the US.

At first, the time series properties of all the pertinent variables are investigated using the appropriate unit root test (Augmented Dickey Fuller) for the period 2001/02Q1—2010/11Q4. The results of the test show that all variables are I(1).20 Initially, a vector autoregression (VAR) is run with all the endogenous variables in their levels for the purpose of deciding

18 In this part of the empirical analysis, the interest rate differential is calculated as the spread between the domestic and foreign interest rates (i – i*). As mentioned earlier in the empirical section, it is the interest rate spread that is responsible for the non-stationarity of the exchange rate-adjusted interest differential (see footnote 15). 19 The seminal Meese and Rogoff (1983) article indicated that fundamentals-based exchange rate determination models (including the Dornbusch model) are not superior to a random walk model for exchange rate determination. Nevertheless, such models are very useful as frameworks of analysis. 20 ADF tests are not presented here, but can be furnished by the author upon request. 20 upon the lag structure. Based on the Akaike information criterion, two lags are selected for all the multivariate time series analyses that appear next.

The long-run relationship between the five endogenous variables included in our model is detected using the Johansen test of cointegration. Indeed, the variables appear to be cointegrated. The Johansen (trace) test for cointegration detects at most 2 cointegration vectors (Table 3).

Table 3. Johansen test for cointegration Unrestricted Cointegration Rank Test (Trace)

Hypothesized Trace 0.05 No. of CE(s) Eigenvalue Statistic Critical Value Prob.**

None * 0.656 84.088 60.061 0.000 At most 1 * 0.462 43.582 40.175 0.021 At most 2 0.280 20.011 24.276 0.157 At most 3 0.165 7.513 12.321 0.277 At most 4 0.018 0.676 4.130 0.471

Note: Trace test indicates 2 cointegrating eqn(s) at the 0.05 level * denotes rejection of the hypothesis at the 0.05 level **MacKinnon-Haug-Michelis (1999) p-values

Unrestricted Cointegration Rank Test (Maximum Eigenvalue)

Hypothesized Max-Eigen 0.05 No. of CE(s) Eigenvalue Statistic Critical Value Prob.**

None * 0.656 40.505 30.440 0.002 At most 1 0.462 23.571 24.159 0.059 At most 2 0.280 12.498 17.797 0.262 At most 3 0.165 6.837 11.225 0.264 At most 4 0.018 0.676 4.130 0.471

Note: Max-eigenvalue test indicates 1 cointegrating eqn(s) at the 0.05 level * denotes rejection of the hypothesis at the 0.05 level **MacKinnon-Haug-Michelis (1999) p-values

We present below the cointegration equation (displaying the interest rate differential on the left-hand side) with the standard errors shown in parenthesis below.

21

= 24.45 + 2.70 ( ) 12.88( ) + 0.63( ) (5) ∗ ∗ ∗ 𝑒𝑒 𝑒𝑒∗ 𝑖𝑖 𝑡𝑡 − 𝑖𝑖 𝑡𝑡 − (11.11) 𝑆𝑆 𝑡𝑡 (8.02) 𝑀𝑀 𝑡𝑡 − 𝑀𝑀 𝑡𝑡 − (7.61) 𝑌𝑌 𝑡𝑡 − 𝑌𝑌 𝑡𝑡 (0.23) 𝜋𝜋 𝑡𝑡 − 𝜋𝜋𝑡𝑡 The objective of displaying the above cointegration equation is to look at the signs and statistical significance of the coefficients of the long-run relationship between the interest rate 21 differential and each of the rest of the endogenous variables.

From the cointegration equation above, we could observe that all variables are statistically significant and appear with the expected signs, with the exception of the monetary aggregate differential.22 We could illustrate what the above expression means from the perspective of the domestic variables that appear in the above cointegration equation. Ceteris paribus, there is a negative association between the domestic interest rate (i) and the bilateral nominal exchange rate (S = LE/US dollar); a negative association between domestic real GDP (Y) and the domestic interest rate (i); and finally a positive association between expected inflation (πe) and the domestic interest rate (i).23 In other words, the above expression means that a depreciation of the foreign value of the Egyptian pound (i.e., an increase in S) is associated with a decrease in the domestic interest rate. That would in turn lead to a decrease in the spread between the domestic and foreign interest rates. Also, monetary tightening (a contractionary policy) which would entail an increase in the domestic interest rate (and a bigger interest rate spread) is associated with a smaller domestic real output; thus the negative association between the real GDP differential and the interest rate spread. Finally, a rise in domestic inflationary expectations is associated with a rise in the domestic interest rate (thus a higher interest rate spread). Having detected the presence of a long-run relationship among the variables, our multivariate time series

21 It is noted, however, that this is not presented as an estimation of the causal relationships. We refrain from presenting this as an estimated interest rate differential because monetary models (including Dornbusch’s sticky- price monetary model which we rely on) have been generally used in the literature to explain the behavior of the exchange rate and not the interest rate differential. Later when the VECM is estimated, the exchange rate is modeled as the dependent variable in the cointegrating vector. 22 Having a statistically insignificant monetary differential may be explained by monetary neutrality. That is, monetary shocks “die out” in the long-run, and thus their effect on the real economy is only transitory. 23 The effect of foreign variables is opposite to that discussed above for the domestic variables. 22 analysis should include an error correction term. Therefore, a vector error correction model 24 (VECM) is estimated.

The short-run dynamics in the VECM are represented in a system of five equations where each endogenous variable is regressed on lagged values of itself and lagged values of all the other endogenous variables, in addition to the error-correction term(s). The error- correction term(s) is basically the lagged error obtained from the cointegration relationship.25 Among the short-run dynamics, the equation that captures the interest rate spread (i - i*) is our main focus.26

Specifically, we would like to investigate how the other four endogenous variables in the model affect the interest rate spread, as well as the relative importance of the various shocks in explaining the variation in the interest rate spread.

In order to do so, we run Granger-causality tests after estimating the VECM. The objective of the Granger-causality test is to see whether the endogenous variables in the model bear predictive content for the interest rate spread. After that, we generate variance decomposition for the interest rate spread in order to identify the major contributor to the variation in it. Granger-causality tests will allow us to interpret the variance decompositions in a “causality” context.

Granger-causality test results for the interest rate differential as the dependent variable are displayed below, while the details are deferred to Appendix 4.

24 The VECM is a restricted form of the vector autoregression (VAR). VECM not only models the joint behavior of the endogenous variables, but also allows for the presence of an additional term that corrects for short-run deviations from the long-run equilibrium. Therefore, in the presence of such a long-run equilibrium relationship, estimating an unconstrained VAR may amount to a mis-specified model, as the short-run deviations are thought to be an important factor that characterizes the relationship between the endogenous variables of our model (Enders 1996). 25 The ordering of the variables in the VECM is important. The variable that comes first is considered to have a contemporaneous effect on itself and all the other variables in the system, whereas the second variable in the ordering is considered to have a contemporaneous effect on itself and the variables that follow it in the ordering, but not on the preceding variable. The VECM was estimated using the following ordering: exchange rate, M2 differential, real GDP differential, inflation differential and interest rate differential. Exchange rate comes first as it is the variable of interest in the Dornbusch model (i.e., it is the explained variable). For the rest of the endogenous variables, the ordering is ad-hoc, and it follows from the way the Dornbusch model equation is presented in various papers that attempted to estimate it starting with a monetary shock (see, for example, Frankel (1984)). And thus, the M2 differential comes as the second endogenous variable in the ordering, as the monetary shock could be thought of as the policy variable in this model; the central bank’s tool to stimulate the economy, and thus would have a contemporaneous effect on the other endogenous variables in the model. It is worth noting that the variance decomposition results (which are our main focus for analysis) do not change drastically for other trials with different orderings. 26 Details of the “interest spread” equation estimated within the VECM are presented in Appendix 3. 23 Table 4. Granger-Causality Test Results

Chi-sq Prob. associated Conclusion Null Hypothesis: statistic with statistic

Reject Ho at 1 percent ∆(Log Ex/Rate) does not Granger-cause ∆(Interest 48.34 0.000 signif. Rate Spread)

Reject Ho at 1 percent ∆(M2 Diff) does not Granger-cause ∆(Interest Rate 28.12 0.000 signif. Spread)

Reject Ho at 5 percent ∆(Y Diff) does not Granger-cause ∆(Interest Rate 7.89 0.0193 signif. Spread)

Reject Ho at 5 percent ∆(Expected Inflation Diff) does not Granger-cause 7.67 0.0216 signif. ∆(Interest Rate Spread)

74.92 0.000 ALL

From the table above we could reject the null hypothesis of no Granger-causality for all the variables, and for the model as a whole. So, we infer that the endogenous variables included in the model are important to the analysis of the interest rate spread. Each variable individually Granger-causes the interest rate spread. This means that lagged values of the exchange rate, M2 differential,27 real GDP differential and the expected inflation differential all help predict the interest rate spread. This finding will thus enable us to interpret the variance decomposition results from a causality perspective. This is what we turn to next.

Table 5. Variance Decomposition of the Interest Rate Spread

Log exchange Real output Expected Interest rate Period S.E. rate M2 DIFF DIFF inflation diff spread 1 1.113 5.013 5.260 0.005 4.409 85.313 2 1.586 6.370 30.131 10.505 2.317 50.677 3 2.195 21.511 36.336 14.182 1.214 26.758 4 2.579 16.165 40.990 18.504 4.183 20.157 5 3.057 12.352 39.558 17.908 14.566 15.616 6 3.449 10.360 36.029 18.007 21.926 13.677 7 3.674 9.175 33.935 17.905 25.770 13.214 8 3.816 8.535 33.192 17.616 27.539 13.118 9 3.949 7.969 33.045 17.781 28.108 13.097 10 4.094 7.469 32.807 18.329 28.378 13.017 From the table above, it seems that the monetary aggregate differential (M2Differential) and the expected inflation differential are the major contributors to the observed variability in

27 Once again, the empirical finding that the monetary aggregate differential (M2diff) Granger-causes the interest rate differential is an indicator of an autonomous monetary policy in Egypt. As mentioned earlier, there are previous studies that argue that monetary policy in Egypt is being complicated by the presence of an open capital account and a stable exchange rate. However, the Granger-causality results further confirm the conclusion that Egypt is only weakly integrated in the world financial market, and that the Central Bank of Egypt is still capable of stimulating changes in monetary variables which in turn affects the interest rate differential, as shown. 24 the interest rate spread. In the beginning, the variation in the interest rate spread is mainly explained by itself. Shocks to the expected inflation differential start to be relatively important as a source of variation in the interest rate spread after 5 forecast periods, and more so towards the 10th forecast period, as more than 28 percent of the variation in the interest rate spread is attributed to shocks to the expected inflation differential. This is a plausible finding as interest rate changes are mainly explained by expected inflation as the Fisher hypothesis postulates.

Shocks to the M2-differental contribute around 33 percent of the variation in the interest rate spread towards the end of the forecast horizon. This result is in line with the finding of the monetary autonomy test, which has shown that interest rate spread is responsive to changes in the domestic monetary aggregate (M2) in the short run.

V. CONCLUSION

The empirical tests conducted in this study proved that Egypt sustains a low degree of (i.e., imperfect) de facto international financial integration: the uncovered interest parity (UIP) does not hold between Egyptian and US financial assets, as evidenced by the non- stationary exchange rate-adjusted differential between the 3-month Treasury bill rates of the two countries. Moreover, it was shown that monetary policy is autonomous: changes in the monetary aggregate (M2) Granger-cause movements in that exchange rate-adjusted differential (i.e., M2 growth creates deviations from the UIP). So, despite the de facto managed exchange rate arrangement that Egypt has adopted since the early 2000s (even after the announced floatation of the exchange rate in January 2003), the Central Bank of Egypt was capable of effecting changes in domestic interest rates. That is to say, even though the exchange rate has been stable, monetary policy in Egypt was still independent in the face of a legally liberalized capital and financial account of the balance of payments.

It is thus said that during the periods under investigation, Egypt has been enjoying de jure financial openness only, with limited de facto integration in the world financial market.

Cross-border capital flows in Egypt could be described as volatile and have not been large enough to wipe out (or at least smoothen) the non-stationary differential that exists between interest rates on domestic and foreign financial assets. While that interest rate differential could be attractive to capital inflows, it seems to constitute a risk factor that deters

25 investors. According to the variance decompositions generated from the VECM, the high inflation rate in Egypt (which implies a large Egy-US inflation differential) has been a main contributor to the variability of that spread between interest rates on domestic and foreign financial assets, and thus could be deemed as a culprit behind Egypt’s limited de facto financial integration, as it led to the persistence of a large and time-varying risk premium/interest rate differential.

Since the early 1990s, Egypt has been facing recurrent episodes of capital inflows with aggressive sterilization; resisting the nominal appreciation of the exchange rate. But Selim (2012b) showed that sterilization was not complete, in the sense that part of the increase in liquidity due to capital inflows was not sterilized, and thus translated into monetary growth and inflation. This gave rise to the vicious cycle of monetary tightening to curb inflation, which meant a further increase in the interest rate differential, attracted capital inflows, and warranted the Central Bank’s intervention to sterilize the effect of the accompanying increase in liquidity. Excess liquidity that remained unsterilized fed into the inflation rate and required further monetary policy tightening, and so on.

Thus, this study points to the lack of policy coordination during the episodes of high capital inflows, which led to overheating in the economy, without reaping the potential gains from the de jure financial openness as the ensuing inflation problem in Egypt was responsible for the persistent large and varying risk premium between the domestic and foreign financial markets.

26

APPENDIX 1. RESTRICTIONS AND ARRANGEMENTS ON CAPITAL: CURRENT STATUS AND DEVELOPMENTS IN THE PAST

Status of Controls on Capital Transaction and Provisions Specific to Financial Developments in the past(as per AREAER for years 1991-2007) Institutions (as per AREAER for years 2008-2009-2010) Repatriation No surrender requirements to the central bank, and no surrender requirements to Prior to 1991, surrender requirements existed. All banks were required to requirements for authorized dealers. deposit 15 percent of their foreign currency deposits with the central bank. capital transactions Proceeds from sales of property owned in Egypt by foreigners (or their heirs) must be deposited in a special capital account in the name of the foreign seller at an authorized bank. See the section “Controls on Real Estate Transactions” below to trace the gradual removal of all types of controls on real estate transactions. Controls on capital No restrictions on nonresidents’ local purchase and sale of capital market Prior to 1994, transactions in Egyptian and foreign securities registered at the market instruments securities (including: shares, other securities of a participating nature, bonds and stock market in Egypt were not transferred through the free market for foreign other debt securities).Also, there are no restrictions on the issuance of such exchange. Instead, imports and exports of securities and transfers related to capital market securities locally by nonresidents. Capital Market Law 95/1992, their purchases or sales were effected through authorized banks. amended by Law 123/2008, allows international institutions to issue bonds in the Since 1994, the foreign exchange market was to be used for transferring local market, subject to approval of the Egyptian Financial Supervisory Authority proceeds associated with the sale of both Egyptian and foreign securities. (EFSA).28 With regards to local sale and/or issue of shares or other securities by However, there are restrictions on residents’ purchase abroad of capital market nonresidents: No controls under the foreign exchange law and regulations on securities. Specifically, private pension funds are not allowed to invest in foreign the issuing of securities in the country by nonresidents. Trading in securities securities or assets abroad. denominated in foreign currency must be settled in foreign currency. Approval of Capital Market Authority (CMA)29 was required for issuing bonds. Prior to June 2008, international institutions were not allowed to issue bonds. June 1st 2008, Law 123/2008 was issued to allow international institutions to issue bonds in the local market, subject to CMA approval.

Controls on money No restrictions on nonresidents’ purchase, sale or issue locally. Also, no market restrictions on residents’ sale or issue abroad. instruments, and But again, purchases abroad by residents of money market instruments and controls on collective investment securities are subject to restrictions as private pension funds collective are not allowed to invest in foreign securities or assets abroad. investment securities

Controls on Derivatives are permitted for purchase, sale and issue by nonresidents in the Prior to 2007, derivatives did not exist in the Egyptian market.

28 The Egyptian Financial Supervisory Authority (EFSA), established by law 10/2009, replaced the Egyptian Insurance Supervisory Authority, the Capital Market Authority, and the Mortgage Finance Authority in application of the provisions of the supervision and regulation of insurance law no. 10 of 1981, the capital market law no. 95 of 1992, the depository and central registry law no. 93 of 2000, the mortgage finance law no. 148 of 2001, as well as other related laws and decrees that are part of the mandates of the above authorities. 29 CMA was later on replaced by EFSA (See previous footnote). 27 derivatives and Egyptian market only for genuine hedging purposes within banks. Also, other instruments derivatives are allowed for sale and issue abroad by residents for hedging purposes only. But purchase abroad by residents is restricted. Controls on credit No controls on commercial or financial credits to residents and nonresidents. But Prior to 2006, there were controls on medium- and long-term commercial operations it is worth noting that for commercial credits from nonresidents to residents, credits from nonresidents to residents. ministries, government agencies, public authorities and private sector entities are No controls were applied if the maturity of the commercial credits is one year all required to register their debt obligations with the CBE. or less and if these credits are received by the private sector. And for financial credits from nonresidents to residents, residents are required to register their foreign debt with the CBE for statistical purposes. Controls on direct No controls on outward direct investment. But with regards to inward direct investment investment, nonbank companies of foreign exchange dealers must be owned entirely by Egyptians. All FDI inflows must be registered by the General Authority for Investment (Laws 8/1997 and 159/1981). Controls on No controls on liquidation of direct investment. Prior to 1997, there were controls on the liquidation of direct investment: liquidation of Specifically, upon the request by foreign investors and subject to the approval direct investment of the board of the General Authority for Investment, repatriation of the invested capital had to be within the limits of its value calculated on the basis of the announced exchange rate at the time of liquidation or disposition. This had to be in 5 equal annual installments, provided that the authority approved the result of the liquidation. Exceptionally, the invested capital was allowed to be transferred in full, if the balance of the foreign currency account held by the investor with one of the authorized bank permits, or if the authority’s board approved making the transfer in full under certain conditions. If the invested capital was done in-kind, it was allowed to be repatriated in-kind. Controls on real No controls on purchase or sale of real estate by residents and nonresidents. There were restrictions that affected the timing of the transfer abroad by estate transactions nonresidents of the proceeds of sales of Egyptian real estate. Such restrictions were eliminated on July 15, 1996. Prior to 2006, local purchase and sale of real estate by nonresidents were subject to controls. From 2003 till end-2005, non-Egyptians were not allowed to sell property within five years after taking possession thereof. All controls on sale and purchase of real estate by nonresidents were abolished in 2006.

Controls on No controls on personal capital transactions (i.e., no controls on personal loans, Prior to 1994, outward personal capital transactions were restricted. Egyptian personal capital gifts, endowments, inheritances and legacies, settlement of debts abroad, transfer emigrants, foreigners leaving Egypt permanently and accrued alimony transactions of assets, transfer of prize earnings). beneficiaries living abroad were authorized to transfer abroad funds up to certain limits prescribed by some pertinent legal provisions. Provisions specific No restrictions on borrowing abroad, nor on maintenance of accounts abroad. July 16, 2003: Law 88/2003 – Banking sector and money law came into effect to commercial No restrictions on lending to nonresidents (financial or commercial credits). requiring that the face value of shares (excluding trading securities) owned by banks No restrictions on lending locally in foreign exchange. banks may not exceed the bank’s capital base. No restrictions on the purchase of locally issued securities denominated in Prior to 1993, regulations governing open foreign currency positions were foreign exchange. different: The difference between a bank’s foreign currency denominated No differential treatment of deposit accounts held by nonresidents (compared to assets and foreign currency denominated liabilities should be limited to

28

that of deposits held by residents). 15percent of the bank’s capital. However, there exists differential treatment of deposit accounts in foreign exchange: Banks’ deposits in foreign currencies (held by Egyptians or foreigners) are subject to a 10 percent reserve requirement, which must be deposited with the CBE and is remunerated at London Interbank Bid Rate (LIBID). Also, there is a requirement for liquid assets in foreign currencies of 25 percent. Egypt applies “fit and proper tests” to shares held by residents and nonresidents in any bank in Egypt that exceed 5 percent of the bank’s issued capital. These require CBE notification. Shares that exceed 10 percent of a bank’s issued capital require CBE approval. There are limits to open foreign exchange positions: the surplus/deficit for each currency may not exceed 10 percent of the capital base. The aggregate surplus/deficit for total long or short positions may not exceed 20 percent of capital base. Provisions specific There are limits (max.) on securities issued by nonresidents for all three types of to institutional institutional investors. investors There are limits (max.) on investment portfolios held abroad for all three types of (1. insurance institutional investors. companies, There are limits (min.) on insurance companies’ and pension funds’ investment 2. pension funds, portfolio held locally. and 3. investment There are regulations pertaining to investment funds’ local and international firms and investments. All investment funds must issue a prospectus approved by the collective EFSA. EFSA’s approval is required for issuance of a public or private offering of investment funds) investment funds.

29 APPENDIX 2. ISSUES PERTAINING TO DATA USED IN THE EMPIRICAL TESTS: The UIP and the monetary autonomy tests were conducted using monthly data. The following are the data sources of each variable: - Interest rate on Egyptian 3-month Treasury bills: Prior to January 2003, series obtained upon request from the Central Bank of Egypt. The rest of the series (January 2003–December 2011) is obtained from the Monthly Statistical Bulletin. - Interest rate on US 13-week Treasury bills: International Financial Statistics online database. - Bilateral exchange rate (LE/$): International Financial Statistics online database. - Domestic liquidity in Egypt (M2): International Financial Statistics online database. The Cointegration/VECM analysis was conducted using quarterly data due to the absence of monthly data for real output in Egypt. - The real output differential was calculated as the difference between the logs of Egypt’s quarterly real output (in LE billions) and that of the US (in USD billions). Egypt's real GDP was obtained from the Ministry of Planning and International Cooperation website: www.mop.gov.eg. - The interest rate differential is calculated as the spread between the Egyptian 3-month Treasury bill rate and the 13-week US Treasury bill rate. - The monetary aggregate differential was calculated as the difference between the logs of Egypt’s domestic liquidity (M2) (in LE billions) and the US counterpart (in USD billions). - The inflation differential was calculated as the difference between the Egyptian inflation rate and the US inflation rate. Both series were obtained from the International Financial Statistics online database. Expected inflation differential was assumed to be the (t+1) series of the inflation differential series. All US data series used to calculate the rest of the differentials were obtained from the International Financial Statistics online database.

30 APPENDIX 3. ESTIMATED VECTOR ERROR CORRECTION MODEL ALLOWING FOR TWO COINTEGRATING EQUATIONS

The short run dynamics of the “interest rate spread equation”, estimated within the VECM.

Sample (adjusted): 2002Q4 – 2012Q1 Included observations: 38 after adjustments

Standard Coefficient Error T-Statistic Cointegration Term 1 2.046 2.889 0.708 Cointegration Term 2 -0.742 0.511 -1.453 DLog Exchange Rate (-1) 0.835 5.112 0.163 DLog Exchange Rate (-2) 34.585 5.387 6.420 DM2_DIFFERENTIAL(-1) -33.863 10.805 -3.134 DM2_DIFFERENTIAL(-2) -34.307 13.697 -2.505 DReal_Output_DIFFERENTIAL(-1) -9.580 3.465 -2.765 DReal_Output_DIFFERENTIAL(-2) -6.744 3.834 -1.759 DINFL_DIFFERENTIAL_Expected(-1) -0.169 0.092 -1.828 DINFL_DIFFERENTIAL_Expected(-2) -0.130 0.106 -1.221 DInterest_Spread(-1) -0.182 0.101 -1.796 DInterest_Spread(-2) -0.242 0.096 -2.528

31 APPENDIX 4. GRANGER-CAUSALITY TEST RESULTS AFTER RUNNING VECM ALLOWING FOR 2 COINTEGRATING VECTORS; AND 2 LAGS

VEC Granger Causality/Block Exogeneity Wald Tests Sample: 2000Q1 2012Q3 Included observations: 38

Dependent variable: D(LOG_EXCHANGE_RATE) Excluded Chi-sq df Prob. D(M2_DIFFERENTIAL) 8.663 2 0.013 D(GDP_ DIFFERENTIAL) 0.255 2 0.880 D(INFL_ DIFFERENTIAL_Expected) 0.828 2 0.661 D(INTEREST_SPREAD) 2.748 2 0.253

All 13.325 8 0.101

Dependent variable: D(M2_DIFFERENTIAL) Excluded Chi-sq df Prob. D(LOG_EXCHANGE_RATE) 0.504 2 0.777 D(GDP_ DIFFERENTIAL) 11.250 2 0.004 D(INFL_ DIFFERENTIAL _Expected) 2.532 2 0.282 D(INTEREST_SPREAD) 0.276 2 0.871 All 16.339 8 0.038

Dependent variable: D(GDP_ DIFFERENTIAL) Excluded Chi-sq df Prob.

D(LOG_EXCHANGE_RATE) 0.510 2 0.775 D(M2DIFFQ2) 4.970 2 0.083 D(INFL_ DIFFERENTIAL _Expected) 0.318 2 0.853 D(INTEREST_SPREAD) 2.767 2 0.251 All 8.401 8 0.395

Dependent variable: D(INFL_DIFFERENTIAL_EXPECTED)

Excluded Chi-sq Df Prob.

D(LOG_EXCHANGE_RATE) 0.411 2 0.814 D(M2_DIFFERENTIAL) 1.676 2 0.433 D(GDP_ DIFFERENTIAL) 2.290 2 0.318 D(INTEREST_SPREAD) 0.924 2 0.630 All 5.392 8 0.715

Dependent variable: D(INTEREST_SPREAD)

Excluded Chi-sq Df Prob.

D(LOG_EXCHANGE_RATE) 48.338 2 0.000 D(M2_DIFFERENTIAL) 28.120 2 0.000 D(GDP_ DIFFERENTIAL) 7.891 2 0.019 D(INFL_DIFFERENTIAL_EXPECTED) 7.674 2 0.022 All 74.915 8 0.000

32 REFERENCES

Abdel-Khalek, Gouda. 2001. Stabilization and Adjustment in Egypt: Reform or Deindustrialization. Cheltenham, UK & Northampton, MA, USA: Edward Elgar.

Aizenman, J., Menzie Chinn and Hiro Ito. 2010. The Emerging Global Financial Architecture: Tracing and Evaluating the New Patterns of the Trilemma's Configurations. Journal of International Money and Finance, 29(4): 615—641.

Al-Mashat, Rania and Andreas Billmeier. 2007. The Monetary Transmission Mechanism in Egypt. International Monetary Fund Working Papers 07/285. Washington, D.C.: IMF.

Alper, C. Emre, Oya Pinar Ardic, and Salih Fendoglu. 2009. The Economics of the Uncovered Interest Parity Condition for Emerging Markets. Journal of Economic Surveys. 23(1): 115–138.

Balli, F., Basher, S.A., Rana, F., 2014. The determinants of the volatility of returns on cross- border asset holdings. Journal of International Money and Finance 44, 1–23.

Bekaert, G. and Campbell Harvey. 1995. Time-Varying World Market Integration. Journal of Finance. 50 (June 1995): 403—444.

Central Bank of Egypt. Various issues. Monthly Statistical Bulletin.

———. Online Time Series: www.cbe.org.eg/timeseries.

Cheung, Yin-Wong, Menzie Chinn and Eiji Fujii. 2006. The Chinese Economies in Global Context: The Integration Process and its Determinants. Journal of the Japanese and International Economies, 20(2006) 128–153.

Chinn, Menzie and Hiro Ito. 2008. A New Measure of Financial Openness. Journal of Comparative Policy Analysis, 10(3): 309–322.

Chinn and Ito Index Online http://web.pdx.edu/~ito/Chinn-Ito_website.htm

Chinn, Menzie and Guy Meredith. 2004. Monetary Policy and Long-Horizon Uncovered Interest Parity. International Monetary Fund Staff Paper.51 (3).

Dornbusch, Rudiger. 1976. Expectations and Exchange Rate Dynamics. Journal of Political Economy 84(6): 1161-76, December. Chicago: University of Chicago Press.

El-Refaie, Faika. 2002. The Coordination of Monetary and Fiscal Policies in Egypt. In Monetary Policy and Exchange Rate Regimes: Options for the Middle East, eds. Ahmed Galal and Eliana Cardoso. Cairo: AUC Press.

Enders, Walter. 1996. RATS Handbook for Applied Econometric Time Series. John Wiley & Sons.

Fleming J. 1962. Domestic Financial Policies under Fixed and under Floating Exchange Rates. International Monetary Fund Staff Papers 9: pp.369-379.

33 Frankel, Jeffrey .1984. Tests of Monetary and Portfolio Balance Models of Exchange Rate Determination. In Exchange Rate Theory and Practice, eds. John Bilson and Richard C. Marston. Chicago: University of Chicago Press. Available online at: http://www.nber.org/chapters/c6837.pdf.

———. 1991. Quantifying International Capital Mobility in the 1980s. In B. Douglas Bernheim and John B. Shoven, eds., National Saving and Economic Performance. Chicago, Ill.: University of Chicago Press.

———. 1996. International Capital Mobility and Exchange Rate Volatility. Federal Reserve, Bank of Boston. Available Online at: http://www.bos.frb.org/economic/conf/conf32/conf32f.pdf.

Hussein, Khaled and Luiz de Mello Jr. 1999. International Capital Mobility in Developing Countries: Theory and evidence. Journal of International Money and Finance 18: 367 – 381.

International Monetary Fund. Various issues. Annual Report on Exchange Arrangements and Exchange Restrictions (AREAER). Washington, D.C.: IMF.

———. 2009. Arab Republic of Egypt – 2009 Article IV Consultation. Washington, D.C.: The International Monetary Fund.

———. 2010. Arab Republic of Egypt – 2010 Article IV Consultation. Washington, D.C.: The International Monetary Fund.

———. International Financial Statistics online database. Washington, D.C.: The International Monetary Fund.

Meese, Richard and Kenneth Rogoff. 1983. The Out-of-Sample Failure of Empirical Exchange Rate Models: Sampling Error or Misspecification? In Exchange Rates and International Macroeconomics, ed. Jacob Frenkel. University of Chicago Press. Chapter available online at: http://www.nber.org/chapters/c11377.pdf.

Ministry of Planning and International Cooperation website: www.mop.gov.eg.

Montiel, Peter. 1989. Money versus Credit in the Determination of Output for Small Open Economies. International Monetary Fund Working Paper WP/89/78. Washington, D.C.:IMF.

———. 1994. Capital Mobility in Developing Countries: Some Measurement Issues and Empirical Estimates. World BankPolicy Research Working Paper Series 1103.Washington, D.C.:The World Bank.

Moursi, Tarek, Mai El-Mossallamy, and Enas Zakareya. 2007. Effect of Some Recent Changes in Egyptian Monetary Policy: Measurement and Evaluation. Egyptian Center for Economic Studies (ECES) Working Paper 122.Cairo, Egypt: ECES.

Mundell, Robert. 1963. Capital Mobility and Stabilization under Fixed and Flexible Exchange Rates. Canadian Journal of Economics and Political Science 29: 459-468.

34 Obstfeld, Maurice, Jay C. Shambaugh, and Alan M. Taylor. 2005. The Trilemma in History: Tradeoffs among Exchange Rates, Monetary Policies, and Capital Mobility. The Review of Economics and Statistics 87(3): 423-438 (August).

Rennhack, Robert and Guillermo Mondino. 1988. Capital Mobility and Monetary Policy in Colombia. International Monetary Fund Working Paper 88/77.Washington, D.C.: IMF.

Selim, Hoda. 2012a. Exploring the Role of the Exchange Rate in Monetary Policy in Egypt. A paper presented in the ERF 18th Annual Conference. March 25-27, 2012. Cairo.

———. 2012b. Has Egypt’s Exchange Rate Policy Changed after the Float? Middle East Development Journal, 4(1).

35